Economics of Education Review 18 1999 361–373
Local public choice of school spending: disaggregating the demand function for educational services
Torberg Falch , Jørn Rattsø
Department of Economics, Norwegian University of Science and Technology, N-7055 Dragvoll, Norway Received 1 January 1997; accepted 5 August 1998
Abstract
The determinants of school spending are investigated using a disaggregated demand model augmented to include political factors. High school spending by county governments is disaggregated to identify the sources of variation in
teacher–student ratio, non-wage spending per student, and student enrollment. The disaggregation throws new light on the role of cost factors in explaining the expansion of educational services. High school spending is shown to be highly
inelastic to county revenue and major cost factors. The spending decision is analyzed as an example of the common pool problem in distributive politics. Schools offer benefits to each municipality, and municipalities fight for new
schools since the costs are shared. The political decision implies a balancing between this spending pressure and the coordinated interests of the county. Political strength, measured by the party fragmentation of the council, is shown to
hold down costs and allow for more student enrollment. On the other hand, the spending pressure measured by the average size of the municipalities in the county, influences all three spending components, and the effects depend on
the political strength.
1999 Elsevier Science Ltd. All rights reserved.
Keywords: Educational finance; Educational economics; Expenditures
1. Introduction
The determination of school spending has been addressed in a large empirical literature recent important
contributions are Craig Inman, 1982, 1986; Romer, Rosenthal Munley, 1992; Rubinfeld Shapiro, 1989.
The analysis of education has been a way of understand- ing public sector decision making, since the education
sector is one of the biggest items in the government budget. The school sector typically is decentralized to
the local public sector, and the cross section variation has allowed empirical investigation of economic deter-
minants.
The standard framework of analysis is the expenditure demand model of education. It is assumed that the local
government organizing school spending is guided by the
Corresponding author. Tel: 1 47-73-596-757; fax: 1 47- 73-596-954; e-mail: torberg.falchsv.ntnu.no
0272-775799 - see front matter
1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 2 - 7 7 5 7 9 8 0 0 0 4 5 - 4
preferences of the voters, usually assuming the decisive role of the median voter. Given preferences and budget
constraints, demand equations are estimated based on cross-section data for school spending, private income
level, tax price and grants. We propose to develop the methodology in two directions. First, school spending is
disaggregated to separate out different components of school expenditure. The disaggregation throws new light
on the role of cost factors in explaining the expansion of educational services. The decomposition identifies three
elements central to the decentralized decision making: teacher–student ratio, non-wage spending per student,
and student enrollment ratio.
Second, the expenditure demand model is augmented to include interest groups and political structure. The
school spending decision is analyzed as an example of the common pool problem in distributive politics
Weingast, Shepsle Johnsen, 1981. The benefits of each school are concentrated to a geographic area, while
the costs are financed by general taxation at a higher
362 T. Falch, J. Rattsø Economics of Education Review 18 1999 361–373
geographic level. Both sides of this equation are included, municipalities fighting for schools in their area
and political leaders at the county level trying to internal- ize the externalities of the school spending decisions.
The empirical basis of our investigation is the vari- ation of high school spending across counties and over
time in Norway. The high school education is an essen- tial part of the welfare state. The national goals emphas-
ize equalization, but the county level freedom of econ- omic priority implies important differences in the
teaching offered. In 1990, the school spending share of the county budget varied from 12 to 24, and spending
per student varied between NOK 32 000 and 52 000 US4500 to 7500. The county with the lowest priority
of high schools enrolled only 64 of its youth, while the highest enrollment rate was 89. Needless to say,
the variation across counties is a source of political dis- cussion. The county decision making must handle the
choice between quality and quantity of high school edu- cation and the priority of high schools against other ser-
vices, notably hospitals.
The centralized system of financing represents another departure from the conventional demand studies emphas-
izing the choice between private consumption and local public services. When local government revenue is
determined by general grants and fixed income tax rev- enue sharing, the local decision making basically allo-
cates a fixed budget between different services. A rationed demand system results, with local government
revenue and service costs as main determinants. A com- prehensive database of economic, political and school
characteristics during the period 1976–93 enables us to do econometric analysis of the determinants of school
spending.
The analysis covers the period after the high school reform of 1976 integrating vocational and academic
training. High schools absorb about 15 of the current expenditures of county governments, as reported in
Table 1. The spending share has been fairly constant on average, first falling and then rising. School spending per
student on average also has been stable in real terms, except for a strong increase during 1990–93. The data
indicate that the high school spending has been driven up by an increasing share of the youth 16–19 years old
enrolled and a rising teacher–student ratio. Close to 90 of the youth is enrolled in high schools in 1993.
1
1
The main motivation of the 1976 Act was to reduce differ- ences in social status between schools of theoretical and occu-
pational orientation. In 1990, the high schools offered more than 100 different specializations. About 65 of the students were
enrolled in the academic track or the commercial and business track, while the rest were enrolled in the vocational tracks com-
prising a large number of specializations such as public health, art, trade, and crafts. They represent higher costs than the tra-
ditional academic track because of the larger need for equip- Table 1
Decomposition of high school spending: mean values School spending
Non-wage School spending
Year share of total
spending per per student
county spending student
1977 0.19 0.030
51 662 6570 13 555 4175
1980 0.16 0.021
53 635 5966 14 437 3578
1985 0.17 0.023
52 155 5865 14 527 3385
1990 0.21 0.025
52 327 5831 13 902 2565
1993 0.23 0.031
57 616 5860 14 388 3908
Teacher– Student
Youth share of Year
student ratio enrollment ratio population
1977 0.085 0.010
0.50 0.070 0.062 0.0031
1980 0.097 0.010
0.57 0.065 0.063 0.0033
1985 0.099 0.011
0.66 0.058 0.067 0.0039
1990 0.102 0.008
0.79 0.071 0.063 0.0026
1993 0.114 0.011
0.89 0.072 0.056 0.0024
Notes: Standard deviations in parentheses. Spending measured in 1993-NOK. NOKUSD 7.
The disaggregated demand model is presented in Sec- tion 2, and the political economy of the decision making
is discussed in Section 3. Following a documentation of operationalization and econometric specification, the
results are presented in Sections 5 and 6, and concluding remarks are offered in Section 7.
2. Disaggregating the demand model of school spending
